Efficiency

Part of public finance theory concentrates on the efficiency effects of taxation, assuming economies are populated by identical representative agents. These theories teach us some important lessons about efficient tax design. By transferring resources from the private to the public sector, taxes inescapably impose a loss on society that goes beyond the revenue generated (except when taxes are “lump sum”). This is because taxes drive a wedge between the price a buyer pays for something and the amount the seller receives. As a result, a tax can prevent some mutually beneficial transactions. For example, a firm will demand labor up to the point where the wage cost (inclusive of tax) equals its marginal product; but a worker will supply labor effort such that its opportunity costs (foregone leisure or home production) equals the after-tax wage. The difference between the wage cost and the after-tax wage is called the labor-tax wedge, which reflects the gap between the value of extra production and the foregone leisure. The reduction in working hours is a pure welfare loss for society over and above the loss from the direct transfer to the public sector. This deadweight loss (or excess burden) is what determines a tax distortion.

Efficient tax design aims to minimize the total deadweight loss of taxes. The size of this loss depends on two main factors. First, losses are bigger the more responsive the tax base is to taxation. Second, the loss increases more than proportionately with the tax rate: adding a distortion to an already high tax rate is more harmful than adding it to a low tax rate. Two prescriptions for efficient tax policy follow: (i) it is efficient to impose taxes at a higher rate if things are in inelastic demand or supply; and (ii) it is best to tax as many things as possible to keep rates low. The latter forms the basis for several policy prescriptions, such as for base broadening and rate reduction.

In the optimum, the deadweight loss of each tax should be equalized at the margin—as otherwise a marginal shift between taxes could reduce the overall tax distortion. Simulations with general equilibrium models show, however, that marginal excess burdens vary significantly between different taxes (Jacobs 2009). This is consistent with empirical studies on the growth impact of taxes, which generally find that income taxes are more distortive for economic growth than taxes on consumption or real property (see e.g. Arnold et al. 2011; Acosta Ormaechea and Yoo, 2012). If we consider the reduction in GDP growth as a proxy for the welfare cost of a tax, this suggests that the tax mix of countries is currently far from efficient.

Equity

Any meaningful tax analysis of equity considerations requires a theory that departs from the representative agent assumption to allow for heterogeneity. Optimal tax theory typically does this by assuming variation in people’s ‘innate ability,’ or ‘talent’. The impact of a tax system on the distribution of after-tax incomes then depends on the progressivity of the tax-benefit system—that is, how rapidly the share of income taken by tax increases with the level of income. Thereby, theory generally allows for the possibility of negative taxes (i.e. transfers). The welfare impact of income redistribution is reflected in the social welfare function, which generally adds the utilities of individuals. Redistribution is desirable for two reasons: first, declining marginal utility of income implies that transferring a dollar from a rich to a poor person will increase the sum of the utilities; second, there is a possibility of social aversion against inequality by assigning higher welfare weights to people with low ability. One extreme here is the pure utilitarian approach that assigns equal weights to individual utilities; the other extreme is the Rawlsian approach that assigns only value to the utility of individuals with the lowest ability.

Income redistribution through a progressive tax-benefit system becomes more complicated if other dimensions are considered. For instance, should redistribution be based on individual or family income? And should progressivity be assessed in terms of annual income—an arbitrary period of measurement—or lifetime income? People may reasonably disagree on these matters and designing an equitable tax-benefit system becomes less straightforward than it may seem. It also raises important related equity issues, such as gender equality and intergenerational equity, which should be taken into account when designing the tax system.

Another important issue for equity is tax incidence. The person who ultimately bears the real burden of a tax may not be the one legally responsible for remitting payment, since taxes can affect market prices (including before-tax wages). The principle is that the burden of a tax— its effective incidence—falls more heavily on the side of the transaction with the least elastic response— that is, the one that finds it more difficult to shift out of the activity being taxed. These price changes as well as general equilibrium effects on the prices in other markets, are often ignored but can matter significantly for the distributional impact of tax policies.

Trade-off between equity and efficiency

Optimal tax theory emphasizes the trade-off between equity and efficiency. Ideally, governments should implement a progressive tax-benefit system that is based on the exogenous innate ability of people.⁶ This would be efficient as it induces no distortions in behavior. Unfortunately, the government cannot observe the talent of people. Instead, it can only observe their income, which is the product of talent and effort (such as in education, training, and work). By setting a tax that is based on income, the tax-benefit system discourages effort and inevitably creates a distortion that is associated with welfare loss. This gives rise to a fundamental trade-off in designing the tax-benefit system, namely between equity and efficiency.

Pioneered by Mirrlees (1971) and advanced by Diamond (1998) and Saez (2001), optimal income tax theory explores how the tax-benefit system can strike an optimal balance between the two. Irrespective of the social welfare weights assumed in the social welfare function, it appears that the average tax burden at the bottom is negative—reflecting transfers provided to those with the lowest income. Then, marginal tax rates should optimally feature a U-shaped form as a function of income – that is, they should be high at the bottom, then fall between the bottom to the middle of the income distribution and subsequently rise from the middle to the top of the distribution. The high marginal tax rate at the bottom is because the transfers being given to the lowest income groups are phased out for middle incomes, as it would be too costly to provide them universally. The low marginal tax rate for the densely populated middle-income groups aims to avoid large aggregate distortions in labor effort. Then, a progressive tax rate structure should be imposed from the middle towards the top of the distribution to increase progressivity. Interestingly, this structure roughly resembles how most systems of means-tested benefits and personal income tax schedules have currently been shaped. ⁷

The assumptions underlying optimal income tax theory have been challenged, however. The view that progressive taxes are bad for efficiency is based on “first-best” analysis in an economy without other, non-fiscal distortions. In practice, however, the world is full of other distortions. For example, markets may be missing (e.g. because of high transaction costs) or fail (externalities, asymmetric information, imperfect competition). In such a “second-best” world, progressive taxes can make a positive contribution to social welfare if they reduce distortions associated with missing and failing markets (yet, they could also exacerbate these distortions). Efficiency improvements have been emphasized in case of imperfections in labor, capital and insurance markets.

• Imperfect labor markets (1): trade unions. Trade unions can exert monopoly power in wage formation by demanding higher wages in negotiations with employers, thus reaping part of the rents from production. This creates structural unemployment because wages exceed the market-clearing level. Although for the trade union it is worth the price of a higher wage, unemployment is socially inefficient. A progressive income tax system can induce trade unions to reduce wage demands because a marginal wage increase produces less net take-home pay for union members. It thus reduces unemployment and increases social welfare.⁸

• Imperfect labor markets (2): frictions. Labor markets are typically characterized by search frictions whereby companies incur costs to fill their vacancies with the right employees and workers must put effort into finding the right job. When they match, workers and firms then negotiate a wage that can exceed the market-clearing wage, depending on the bargaining power of the worker versus the employer. The higher wage, however, increases unemployment. Progressive taxes can encourage employees to settle at lower wages and boost vacancies so that unemployment falls.

• Imperfect capital markets (1): borrowing. People might not be able to borrow the amounts they prefer when young due to moral hazard (behavior that reduces the probability they pay back a loan) or legal restrictions (e.g. a bank cannot force people to work to pay back a loan). Progressive taxes can redistribute income from periods when income is high (higher working age) to periods when income is low (young age) and thus improve how people can spread their consumption optimally over their lives. They thus mitigate adverse effects of capital market imperfections.

• Imperfect capital markets (2): learning. Young people who are rationed on the capital market (e.g. students of poor parents) are not given the opportunity to invest in education. This is inefficient because potentially profitable human capital is not being built. Progressive taxes can again be efficient because they redistribute income over the life cycle, from old to young. Moreover, to the extent that income is redistributed from rich parents to poor parents, adverse effects of failing capital markets can be further limited.

• Imperfect insurance markets (1): job risk. Insurance markets fail due to asymmetric information, leading to adverse selection and moral hazard. Without insurance, people will reduce risk taking, e.g. by avoiding relatively risky jobs. Progressive taxes can mitigate this distortion by redistributing income from those with luck (i.e. a high income) to those with misfortune (i.e. a low income) – which effectively serves as an insurance device. More risk taking improves efficiency.

• Imperfect insurance markets (2): learning risks. As it is impossible to obtain insurance against the risk of returns to schooling, risk-averse people will invest too little in their education. Progressive taxes can serve as an insurance device and encourage such investments in knowledge.

These market imperfections thus qualify the presumed disruptive influence of progressive taxes on efficiency. Empirical evidence supports these theories, e.g. high marginal tax rates are found to reduce wages and structural unemployment (Sorensen 1997). Gerber et al. (2018) find that tax progressivity exerts no adverse effect on economic growth.

Enforcement

A critical element for equity and efficiency is to minimize both tax avoidance (legal) and evasion (illegal). The dividing line between them is not as clear-cut as it may sound, but both are major concerns for governments. Tax avoidance should be addressed by good tax design, e.g. by minimizing opportunities for tax arbitrage such as income shifting or by imposing tight anti-avoidance measures. Minimizing tax evasion requires a good implementation and enforcement by the tax administration. This can be facilitated by making things simple for taxpayers who want to comply through self-assessment. Ultimately, tax administrations should ensure that the probability of detecting noncompliance—and the penalties that follow—is high enough to encourage compliance while supporting and reflecting widespread willingness to follow the rules.

There are special concerns on tax compliance in developing countries, where constraints on the implementation and enforcement of taxes are often more binding due to limited capacity. Moreover, underdeveloped capital markets and a large fraction of the population in informal self-employment render the opportunities for tax design more limited. In fact, constraints have often kept aggregate tax ratios very low. Simplicity in tax law design, easy tax collection from a limited number of sources, and ample self-assessment is therefore important to deal with these enforcement constraints and provide a rationale for e.g. corporate income taxes (especially to collect tax from retained corporate earnings that are otherwise hard to tax), value-added tax (for which the self-enforcing mechanism provides incentives for voluntary compliance), withholding taxes (exploiting administrative powers of large corporations and banks to reduce collection costs), excises (which can often be levied from just a few large businesses, such as breweries, tobacco companies and oil companies) and import tariffs (which can easily be collected on physical goods at the border).

The rest of this paper links the lessons from tax theory to the practice of tax policy making, focusing on the design of various taxes in support of inclusive economic growth.

Use individual income as the tax unit

The unit of income taxation can be either family or individual income. Traditional family-based taxation is often based on either family quotients to determine the joint tax liability (e.g. in France) or income splitting (e.g. in Germany). The latter means that incomes of both partners in a couple are first aggregated and then split in two equal halves that are taxed at the prevailing progressive rate structure. Joint filing has the attraction that it is neutral with respect to the choices made within the household, as the tax due for the family does not depend on which partner generates the income. However, if the incomes of two partners vary (perhaps being zero for one of them) they could significantly reduce their joint tax liability by marrying and filing jointly (a ‘marriage bonus’).⁹ Family taxation could thus distort the choice of cohabitation. Moreover, in a progressive tax system, income splitting reduces the marginal tax rate of the primary earner (the partner with the highest income) and increases it for the secondary earner (the partner with the lowest income). Since the latter are often women, family-based tax systems are typically not gender neutral but instead disadvantage women by imposing a higher marginal tax burden on their income. As the elasticity of labor supply is generally found to be considerably higher for women than for men, family-based tax systems also discourage overall labor supply. For instance, Jaumotte (2004) provides a comprehensive set of simulations for OECD countries, based on estimated aggregate labor-force participation equations, and finds that eliminating tax discrimination against secondary earners (relative to singles) would raise the labor-force participation rate of women by 3.9 percentage points, on average across countries.

Many advanced economies have transformed their PIT into an individualized system. Fully individualized PIT systems are now in place in most European countries following initial reforms in the 1970s and 1980s in Scandinavia, Austria, the Netherlands and in 1990 in the U.K. Other countries with predominantly individualized systems have maintained some elements of family-based taxation, such as a transferable tax deduction from the non-working spouse to the breadwinner, family-based deductions, dependent spouse deductions, or options for joint filing. Countries have also introduced other features to improve the labor market participation of secondary earners, such as tax credits or deductions for childcare expenses or targeted deductions for the income of secondary earners.

Choose an appropriate PIT threshold

A threshold—either in the form of a zero-tax bracket, a basic deduction or a general tax credit¹⁰—supports tax progression by reducing or eliminating the tax burden on people with the lowest incomes. Thresholds vary significantly across economies. In the OECD, the median is approximately 25 percent of the average wage. Several emerging and developing economies, however, have no threshold at all and introducing one could relieve the poorest households from the obligation to pay tax and ease administration. However, the threshold should not be too high either, as this can lead to greatly reduced revenues. In several developing economies, for instance, the threshold exceeds two times GDP per capita (Figure 3). This leads to a very small coverage of the PIT and a low revenue yield, thus undermining redistributive income taxation.

View Full Size

Personal Income Tax Threshold, 2019 (Percent of GDP per capita)

Citation: IMF Working Papers 2020, 271; 10.5089/9781513561561.001.A001

Source: IMF staff calculations.LIDCs = Low -Income a nd Devel oping Countri es (57)EMs = Emerging Markets (96)

• Download Figure
• Download figure as PowerPoint slide

Provide relief for low-income wage earners

Optimal income tax theory provides a rationale for subsidizing earnings of low-wage workers, both because they are relatively responsive to tax (so that relief will eliminate severe inefficiencies) and because it contributes to reducing inequality. In many advanced economies,¹¹ these take the form of refundable tax credits, which constitute a net transfer to the individual when they exceed income tax liabilities. These benefits increase the net income gain from accepting a job relative to the alternative of being out of work and provide income support. In-work benefits are usually phased out as incomes rise, with the steepness of phase-out depending on the primary objective of the program. In countries that emphasize the labor force participation objective, benefits are usually gradually phased out with individual income (Belgium, Finland, Germany, the Netherlands, and Sweden).¹² In countries that emphasize the income support objective, benefits are often conditional on the presence of children in the household and generally phased out more steeply with family income so as to prevent leakage of benefits to higher income families and reduce fiscal costs (Canada, France, Korea, New Zealand, the Slovak Republic, the United Kingdom, and the United States). In the U.S. in 2017, 27 million eligible workers and families nationwide received about $65 billion in tax credits under the earned income tax credit. The federal government estimates that it lifted 9.4 million people out of poverty, including 5 million children.¹³

Rationalize tax deductions

Many countries—including developing ones—adopt various tax allowances in the PIT related to children, education, housing, health insurance, commuting and charitable donations. Some accrue disproportionately to the rich, such as deductions for mortgage interest, because households with high incomes are more often homeowners. Rationalizing mortgage interest deductibility could complement steps towards a more progressive tax system and improve efficiency, since these deductions create their own distortions.

More generally, PIT tax expenditures (defined as specific provisions in the tax code that allow certain people or companies to pay less tax) in several countries accrue disproportionally to people with high incomes. For example, Toder and Baneman (2012) find that if all individual income tax expenditures in the U.S. had been eliminated in 2011, the outcome would have been broadly progressive, with a 19.8 percent decline in after-tax income for the top 1 percent of the income distribution, compared to only a 7.5 percent decrease for the bottom quintile, with additional revenue available for pro-inclusive tax and expenditure changes. In the same vein, the Committee for Economic Development of Australia (CEDA, 2015) finds that superannuation¹⁴ in Australia carries tax concessions that primarily benefit the rich, with the top 20 percent of income earners receiving almost 60 percent of tax expenditures.

Use a stepwise rising PIT rate schedule

Flat PIT schedules often have appeal to policy makers as they signal simplicity and efficiency. Both arguments are flawed, however. Simplicity is a matter of designing the tax base, e.g. by minimizing deductions, credits and exemptions. The rates that are applied to this tax base will then be a simple calculation. Efficiency, as we have seen above, requires a nonlinear structure of the PIT with rising marginal tax rates towards the top—not a flat tax. This shape is independent of the social aversion against inequality.

Nevertheless, since the mid-1990s 27 countries—especially in Central and Eastern Europe and Central Asia—have introduced flat PIT systems, usually with a low marginal rate (Figure 4). When combined with a threshold, these schemes are still progressive in the sense that the average tax burden rises with income. However, significantly more progression can be achieved by using a piecewise linear tax system whereby marginal tax rates increase with income. The challenge is to find a tax rate schedule that strikes a proper balance between equity and efficiency.¹⁵

View Full Size

Countries with Flat PIT Rates, 2019 (Percent)

Citation: IMF Working Papers 2020, 271; 10.5089/9781513561561.001.A001

Source: IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

Set the top PIT rate at an appropriate level

Income tax progressivity has declined steeply in the 1980s and 1990s and has remained broadly stable since (IMF 2017). For instance, the median top PIT rate (across various country groups) dropped from 47 percent in 1990 to 30 percent in the past decade (Figure 5).

View Full Size

Average Top PIT Rates, 1990–2019 (Percent)

Citation: IMF Working Papers 2020, 271; 10.5089/9781513561561.001.A001

Source: IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

The desirable top PIT tax rate is a contentious policy issue, but optimal tax models provide some guidance. For instance, if one assigns zero welfare weight to top income earners (Rawlsian social welfare), it would be optimal simply to maximize revenue collected from them. This requires balancing the revenue gain from a higher marginal top PIT rate at the initial base against the revenue loss induced by behavioral responses that a higher tax rate would induce—such as reduced labor effort, avoidance or evasion—measured by the elasticity of taxable income (see Saez and others 2012 for a review). Studies estimating this revenue-maximizing rate find that it generally ranges between 50 and 60 percent (IMF, 2013b).¹⁶ Some argue that it may be even higher—up to 80 percent—as higher top rates can help discourage rent seeking by top-income earners (e.g. managers who might be able to partly set their own pay by bargaining harder or influencing compensation committees) (Piketty et al 2014). Others, however, emphasize that high marginal rates cause other adverse economic effects, e.g. on innovation and entrepreneurship, and thus create larger economic costs than is sometimes assumed (Akcigit et al 2019). Moreover, the calculations rely on the extreme assumption of a zero welfare weight for the very rich. If a positive welfare weight is assigned to them, the optimal top PIT rate will be lower. Yet, some countries seem to have room to increase their top PIT rate to boost revenue and strengthen progressivity.

Ensure neutral taxation of entrepreneurial income

One reason why a tax on capital income is important is because it is a necessary backstop in the taxation of entrepreneurial income. Indeed, it is often difficult (or even impossible) for tax administrations to distinguish labor income from capital income earned by self-employed entrepreneurs. Businesses organized as sole proprietorships are therefore generally taxed on their total income through the prevailing PIT—and no distinction is made between the labor and capital components of that income. However, entrepreneurs might also opt to run their business as a closely-held corporation. The business then pays its owner-director some fixed remuneration for its labor, which would be subject to the PIT scheme. The remainder of its business income would be seen as capital income and taxed under the corporate income tax (CIT), possibly complemented by a tax on dividend distributions. However, if there is a large difference between the PIT and the CIT treatments, these entrepreneurs will have an incentive to manipulate the share of labor and capital income so as to minimize their overall tax liability. As this will be hard to verify by the government, such arbitrage looms large. To avoid this, it is important to broadly harmonize the rates of the PIT and the combined burden of the CIT and dividend taxation.

Tax different types of investment income uniformly

In many countries, different forms of investment income are taxed at different tax rates. For instance, interest payments are usually deductible for the CIT, whereas returns on equity are not. As a result, interest received by individuals is more lightly taxed than equity returns. At the personal level, moreover, dividends are often taxed at much higher rates than capital gains—which are sometimes left entirely untaxed. This induces firms to retain earnings in the firm or distribute profit by buying back shares. Hence, neutrality of capital income taxation in many countries calls for higher taxes interest and capital gains as compared to dividends.

Many countries also have some form of preferential treatment for certain investors or investments by exempting them from PIT. For example, capital returns of pension funds are often untaxed; and some countries exempt interest on government bonds or other types of returns. All these differences induce changes in asset portfolios of investors that erode the capital income tax base and create economic distortions. A neutral, uniform treatment of all investment income without exemptions is the best way to mitigate this.

Minimize evasion

Taxing capital income at the individual level can be administratively challenging because people can often hide their income from the tax administration, e.g. by holding their assets abroad. The key challenge for tax administrations is to collect and use third-party information to implement taxes on capital income. For instance, the implementation challenges provide a very strong rationale for withholding taxes at the level of the firm, that is, through the CIT. Moreover, withholding taxes on interest and dividends can, to some extent, further circumvent administrative difficulties as they utilize banks and large corporations to collect taxes. Some Latin American countries also impose withholding taxes on capital gains.

An important development for the enforcement of capital income taxes is the increasing prevalence of arrangements to exchange information between countries for tax purposes. Especially automatic exchange of information (AEOI) under the initiative of the G20 and the U.S. Foreign Account Tax Compliance Act (FATCA) hold the prospect of facilitating the enforcement of both capital income taxes and net wealth taxes. A recent IMF study by Beer et al. (2019) finds that the introduction of these standards has already reduced deposits held in low-tax jurisdictions by 25 percent—indicating their effectiveness in curbing tax evasion. Global adoption of these standards is necessary, however, as otherwise deposits might be shifted elsewhere where they can remain hidden. Moreover, for developing countries, there remains a challenge to effectively utilize such information to enforce the taxation of capital income.

Strengthen recurrent property taxes¹⁹

Real property taxes are imposed on gross values. They are among the least distortive for economic growth as their base is immobile. Given that they are paid mainly by residents, and property values likely reflect the value of local public services, property taxes can resemble a benefit tax which can support accountability of local authorities. To the extent that property taxes are capitalized into house prices (e.g. if their supply is fixed), they can also be somewhat progressive since home values generally increase with income and net total wealth holdings.²⁰

Property taxes raise on average around 1 percent of GDP in advanced economies, although their yield goes up to 3 percent of GDP in the UK and Canada. In developing countries, they generally raise less than ½ percent of GDP. In many countries, there is scope to exploit this tax more fully by raising tax rates, updating property values to current market prices and, especially in developing countries, improving cadasters and scaling up administrative capacity. Where market-based valuation is hard, simplified approaches based on property areas can produce reasonable outcomes at lower administrative costs. Property taxes can also be imposed on their transfers. Property transfer taxes may be easier to collect at the time of a sale, which has particular appeal in a low-capacity environment. Yet, these taxes are less efficient than recurrent property taxes, as they reduce mutually beneficial transactions. In some countries (e.g. Belgium, Italy and South Korea) property transfer taxes raise over 1 percent of GDP. A shift toward recurrent property taxation could mitigate distortions in housing markets.

The challenge of a net wealth tax (NWT)

When considering NWTs, it is important to recognize that for assets that generate a return, there are important similarities between a tax on that return and a tax on the underlying asset. This may imply that governments should choose between either a capital income tax or a NWT. Economists generally favor the former for several reasons:

• ■ Equity. If wealthy households generate higher rates of return, e.g., due to higher risk-taking, the effective tax burden on capital income will be more progressive under a capital income tax than under a NWT.

• ■ Efficiency. Like a capital income tax, a NWT affects the after-tax returns to savings, thereby distorting capital accumulation and work effort. Yet, the government does not share in the investment risk under a NWT, unlike under an income tax with loss offset. This means that NWTs distort innovation and entrepreneurship more under a NWT and thus hurt (long-term) economic growth.

• ■ Administration. Capital income taxes might be easier to enforce as tax can be withheld at source, e.g. on interest or dividend payments. However, similar mechanisms could in principle be applied to the stock of wealth.

The similarity above applies to financial assets that generate a rate of return. However, wealth also comprises of non-financial assets such as immovable property and other valuables that do not generate a flow of income (e.g. cars, boats, jewelry, art, etc). A comprehensive NWT would thus have a wider base than a tax on capital income. This wider base, however, might be difficult to enforce.

The key to enforcing a NWT is information reporting by third parties. For instance, financial institutions should provide end-of-year wealth balances for interest-bearing assets, publicly-listed stocks, assets held by mutual funds, and mortgage or student loans; pension funds should report the value of individual retirement accounts; and local governments should share information about the value of real estate and registered vehicles. However, implementation issues may arise with the valuation of certain non-financial assets. For instance, difficulties generally arise with the valuation of closely-held businesses which are not traded, defined benefit pension assets, and works of art and other valuables. Such assets have therefore often been excluded from NWTs, which then subsequently induces scope for tax avoidance. Evasion by the wealthy has been particularly difficult to uncover through traditional means, such as random audits and self-reporting. Hence, the introduction of a NWT will require increased resources for enforcement, reporting by third parties, and arrangements to exchange information between countries.²¹ These issues likely make a NWT less feasible for countries with weak administrative capacity.

The value of wealth held offshore is also hard to ascertain, as information is more difficult or even impossible to obtain. Zucman (2013) estimates the portion of global financial wealth that is held in offshore low-tax jurisdictions at 8 percent—much of which is likely to go untaxed. Studies on Scandinavia suggest that such offshore holdings are most prevalent among the ultra-wealthy (i.e., the top 0.1 percent of the wealth distribution). As discussed above, these enforcement challenges of a NWT might be mitigated by the increasing prevalence of arrangements to exchange information between countries for tax purposes.

A handful of studies have estimated the extent to which NWTs induce behavioral responses in form of reduced saving/investment, avoidance behavior (i.e. legal shifts toward exempt or tax-preferred assets) and tax evasion (illegal misreporting of wealth) (Scheuer and Slemrod 2020). This is reflected in the tax elasticity of reported wealth. In Denmark, Spain and Switzerland effects are estimated to be very large, in order of a decline in reported wealth between 32 and 44 percent in response to a 1 percent tax on wealth. Using evidence from the Panama papers, it appears that a significant portion of these behavioral responses is due to evasion; and this response tends to increase with the level of wealth, i.e. evasion is much larger for the ultra-wealthy (Alstadsæter et al. 2019).

The prevalence of comprehensive NWTs has declined over time. Several countries that had a NWT in the 1990s, for instance, have repealed it (e.g., Australia, Canada, Pakistan and several European countries, including recently France)—in some cases due to limited revenue relative to administrative effort or for constitutional reasons. However, NWTs have survived in Italy, Norway and Switzerland and have been reinstated by Belgium and Spain. In Switzerland, all cantons levy a NWT based on the balance of the worldwide gross assets minus debts. This tax yields around 1 percent of GDP. The Netherlands does not tax capital income but instead taxes a presumed return on savings and investments—referred to as the assets in “Box 3”—comparable to a NWT. It raises around 0.6 percent of GDP. In other countries, revenue from NWTs has been (much) lower.

Tax inheritances and gifts

Inheritance taxes could play a useful role in limiting inter-generational inequality and strengthening equality of opportunity.²² A distinction can be made between intentional and unintentional bequests. The latter are most efficient as they do not affect behavior. The former may lead to avoidance behavior, e.g. people can transfer gifts during life. Therefore, countries usually supplement inheritance taxes with gift taxes. Inheritance and gift taxes might be easier to enforce than NWTs as the transfer of wealth avoids liquidity problems that could otherwise arise. However, complex tax structures with ample forms of preferential treatment and exemptions have often facilitated tax avoidance. As a result, they have not proved easy to implement. In 9 OECD countries, inheritance taxes no longer exist; and where they still exist, rates of tax are generally low with widespread exemptions and ample avoidance. Revenues are on average around 0.1 percent of GDP. There seems potential to increase revenues from them, however: in Belgium and France, for example, they raise 0.7 and 0.6 percent of GDP, respectively.

Design the CIT as a rent tax

The key drawback of how CIT systems are currently designed in most countries is that they create two major economic distortions. First, by raising the cost of capital on equity they distort investment decisions. This hurts economic growth and adversely affects efficiency. Second, by differentiating between debt and equity, they induce a bias toward debt finance. This not only creates an additional direct welfare loss, but also threatens financial stability.²⁵

Both distortions can be eliminated by designing the CIT as a rent tax.²⁶ There are different ways to doing this. First, one class of rent taxes is known as cash-flow taxes, which allow for full expensing of investment instead of deductions for tax depreciation. The simplest form is a real-base cash-flow tax, defined as the net sum of all real receipts and payments, excluding financial flows such as interest payments, net debt issuance, and net dividends. A second type of cash-flow tax is levied on real and financial cash-flows, which adds inflows from issuing loans or interest received and deducts outflows, such as repayments and interest costs. In practice, pure cash-flow taxes are rare. However, there are many examples of countries implementing some of their features. Some countries, for example, have temporarily allowed expensing of investment, but without restricting interest deductibility, such as the US does since its latest reform in 2017. Other countries use cash-flow tax features on surtaxes, for example in the natural resource sector, to capture resource rents (IMF 2012). Mexico for some time had a cash-flow tax that served as an alternative minimum tax.

An alternative rent-based tax system keeps the current CIT but adds a deduction for a notional return on corporate equity—to equalize the treatment of interest. Similarly, the so-called allowance for corporate capital (ACC) replaces the deductibility of actual interest with a notional interest rate applied on all capital, i.e. debt and equity, to obtain neutrality. Allowance for corporate equity systems have been implemented, sometimes only for a few years, in a several countries (Belgium, Croatia, Cyprus, Italy, Latvia, Liechtenstein, Malta, Turkey). Studies generally find powerful effects of the introduction of these systems on corporate debt levels; and some studies also report positive effects on investment. Simulations with a general equilibrium model suggests that the introduction of an allowance for corporate equity in Europe will reduce the corporate debt-to-asset share by 5 percent, boost investment by around 6 percent and raise GDP by 2 percent (De Mooij and Devereux 2011). ACC-like systems were in place in several countries during the world wars, known as excess profits taxes, to finance extraordinary spending.

Consider cost-based rather than profit-based tax incentives

Most countries use some form of tax incentives to mitigate the distortionary impact of the CIT on investment. In advanced economies, especially investments in research and development (R&D) are incentivized. In developing countries, incentives often focus on attracting foreign direct investment (FDI). For both, their design can often be improved by focusing the incentives on reducing the cost of investment, rather than on providing tax relief on profits.²⁷

Many advanced countries encourage R&D by special tax credits or super deductions. These policies can be efficient due to the positive externalities associated with R&D. Empirical evidence suggests that these policies have worked well in many countries and that they hold the prospect of generating significant positive long-term growth effects. The April 2016 Fiscal Monitor, for example, finds that efficient pricing of R&D can boost GDP by 5 percent on average. Recently, some countries have adopted a different type of incentive that does not focus on reducing the costs of R&D but rather on reducing the tax on the profits they generate. These so-called patent box regimes have become particularly prevalent in Europe. Evaluation studies indicate that these regimes either had no discernible impact on R&D or, where they did have an impact, had significant fiscal costs. Indeed, incentives that reduce the costs of R&D investments directly are more cost-efficient than profit-based measures.

Many developing countries aim to attract FDI by providing outright tax exemptions, e.g. in special economic zones, or through time-bound tax holidays. However, these incentives are generally found to be ineffective and inefficient. Indeed, their fiscal cost can be high, while surveys show that these tax incentives generally rank low in the list of relevant location factors for multinationals. Surveys also indicate that profit-based tax incentives are often redundant—that is, the investment would have been undertaken also without them. Investment tax incentives that directly reduce the cost of investment, such as investment tax credits, accelerated depreciation or outright expensing of investment yield more investment per dollar spent.²⁸

Adopt tailored anti-tax avoidance measures

A major risk for the CIT base of countries is due to profit shifting by multinational companies. For instance, international businesses can use transfer pricing techniques, international debt shifting and treaty shopping to reduce their tax liability in a country (see Beer et al. 2020 for a review of evidence on profit shifting). To protect the CIT base against profit shifting, it is important for countries to adopt appropriate anti-tax avoidance measures. The OECD project on base erosion and profit shifting (BEPS) agreed upon common rules and guidance for countries to do so. The European Union has adopted two Anti-Tax Avoidance Directives, to make some of these outcomes mandatory for its member states. In the US, the Tax Cuts and Jobs Act contains anti-tax avoidance provisions that go beyond the BEPS standards by introducing minimum taxes on outbound and inbound investment. In developing countries, anti-tax avoidance measures are particularly important, as these countries suffer relatively more from profit shifting than advanced economies (Crivelli et al. 2016). The capacity to enforce anti-avoidance measures in these countries is constrained, however, due to limited administrative capacity. Moreover, developing countries face distinct problems associated with taxing income at source. To address profit shifting and strengthen the revenue base, developing countries require tailored solutions, for instance in the form of simple caps on CIT deductions to foreign related companies or the adoption of alternative minimum taxes, imposed on assets, turnover or cash-flow (Aslam and Coelho, 2020).²⁹

Design an efficient VAT³⁰

Today, more than 160 countries have a VAT system in place, which resembles a broad-based consumption tax. These systems typically account for around one-quarter or more of total tax revenue. A VAT is imposed on every transaction in the production process. When goods are purchased by a VAT registered business, the tax paid on inputs is credited or refunded. This ensures that VAT is ultimately only levied on final consumption, where no credits or refunds are provided. This design has the attraction of encouraging voluntary tax compliance, since each business has an incentive to register in order to claim credits on their inputs. Moreover, those who operate in the informal sector are still charged on their inputs, without them being able to claim credits. Nevertheless, VAT design in many countries can be full of exemptions and differential rates, which reduces these advantages and induces economic distortions. Acosta Ormaechea and Morozumi (2019) find that this can be damaging for economic growth. To effectively raise revenue from VAT while doing the least damage to inclusive growth, VAT systems can best be designed with a high threshold, a broad base, and a single rate.

• A sufficiently high threshold aims to reduce the compliance costs of VAT for small traders. At the same time, the revenue foregone can be minimal, in part because unregistered businesses cannot claim input credits. A VAT threshold can also strengthen the progressivity of the VAT by reducing the tax on small traders in rural areas where the poorest often buy their goods (Jenkins, Jenkins, and Kuo, 2006).

• Minimize VAT exemptions. Some countries exempt basic goods from VAT to mitigate its distributional effects. However, exemptions are inefficient to achieve a more equitable outcome. First, they cause distortions due to cascading effects (tax on tax) if applied to intermediate stages of the supply chain, as exempt businesses cannot claim credits on their inputs. Second, exemptions create a bias against outsourcing by businesses, since the tax burden can be reduced by producing inputs in-house rather than purchasing taxable inputs from third parties. This creates another inefficiency. Third, by exempting suppliers, the incentives for voluntary compliance with the VAT are reduced. A limited number of well-defined exemptions is quite common, however, for practical reasons. Standard exemptions are applied for example for margin-based financial services, basic health care and education.

• Use a single VAT rate. Some countries tax necessities such as food and medicine at special reduced VAT rates to pursue redistributive policies. The idea is that the poor spend a large proportion of their income on them so that a reduced or even zero rate³¹ offers relief. However, this policy is inferior to pursuing redistribution through other tax and spending policies since it is poorly targeted. For instance, as the rich spend a larger absolute amount on such goods, a large portion of the benefits from a reduced rate accrue to them. Moreover, differential rates significantly complicate VAT administration and cause complexity in defining what goods precisely fall under the reduced rate. Spending measures, such as transfers in cash or in kind, are considerably more efficient to achieve distributional objectives than reduced VAT rates. In developing countries, where the availability of these spending instruments is less common, some reduced VAT rates can be justified on equity grounds—albeit at a high revenue cost.³²

Impose environmental taxes

A key efficiency reason for differential consumption taxation is due to externalities. They arise if the consumption of a good affects the wellbeing of those not involved in the underlying transaction. Environmental damage, such as climate change, is the leading example of a negative externality. The corrective tax, also called a Pigouvian tax, is designed to internalize the external cost of a transaction in the price (i.e. “setting the price right”) so that agents change their behavior in the desired direction, e.g. by reducing pollution.³³ Pigouvian taxes generally come on top of broad-based consumption taxes and should be imposed per unit of consumption, at a level directly related to the external costs—such as the social cost of carbon emissions in the case of climate externalities.

As environmental taxes can be regressive, their introduction may require offsetting tax or spending measures to compensate poor households. In advanced economies, environmental taxes (including on energy) raise around 1.5 percent of GDP in revenue; in developing countries this is usually lower. In many countries, there is significant potential for these taxes to yield more revenue while also improving environmental quality. For example, in the EU, environmental taxes raise around 2.5 percent of GDP. This seems to have had no negative impact on economic growth. For instance, a recent study by Metcalf and Stock (2020) finds no evidence for negative growth effects of carbon taxes in Europe.³⁴ A carbon tax of $75 per ton, necessary to meet the Paris climate objectives, has been estimated to yield more than 1.5 percent of GDP in G20 countries.

Use specific excises

Excises on alcohol, tobacco and unhealthy food (’sin goods’) are generally motivated by related social concerns—although not strictly speaking externalities. Bounded rationality of households and lack of self-control may justify government intervention in the pricing of these addictive commodities—although such arguments are not undisputed. Most countries use excises on these products as part of their policy to improve health outcomes, but revenue raising objectives are important as well. For developing countries, these excises can have special appeal as concentrated production and high import shares make administration relatively easy. Revenue from excises (including on fuel products) varies from an average of 1 percent of GDP in low-income countries to around 2.5 percent of GDP in advanced economies. Over time, revenue has often declined in several countries due to a lack of indexation of the specific (i.e. per unit) rates, which causes revenue to fall with inflation. In many countries, there is scope to raise significantly more revenue from excises without adverse distributional effects (Cnossen 2020). Indeed, while excises tend to bear relatively more heavily on people with low incomes, this only holds in advanced economies and not for developing countries. Special excises on luxury goods, such as yachts, jewelry or perfumes usually contribute little to achieving equity objectives, raise little revenue, and add to administrative costs. The exception is excises on motor vehicles, which can raise sizable amounts and are generally progressive.

Ensure strong political commitment and leadership

Tax reform affects all factions of society and will need the support from the entire government—calling for a “whole-of-government approach”. The Minister of Finance is usually responsible for the management of the tax system and should have the clear and unconditional leadership of the reform effort. In many countries, a permanent or temporary committee chaired by the Minister of Finance and supported by technical working groups, brings together all stakeholders from the public sector to reflect the different interests.

Build consensus and generate public support

Gaspar et al. (2016b) find that, aside from credible leadership, inclusive politics and constitutive institutions are vital elements of tax capacity building. Indeed, it is essential to hold extensive political consultation with multiple stakeholders in society, such as businesses, tax professionals, civil society organizations, local governments, academic researchers, think tanks, etc. Consultative discussions might not create unanimous approval or support, but will instill in society a sense of country-wide ownership. Some groups in society might not be well-organized or integrated into the formal economy—often low-income groups that could benefit most from measures to support inclusion. This may require special attention from the government. Other groups can be very well organized and could even become a stumbling block to the reform if they aim to protect vested interests—often the more affluent who can e.g. oppose increases in tax progression. Managing these political differences requires forceful leadership and strong communication.

Develop a clear and broad communication strategy

The communication campaign should develop a narrative to position the tax reform as a government-led and country-owned strategy that aims to support inclusive growth, emphasizing the benefits to society at large. The government should mobilize representatives from the public sector, the private sector, business associations, religious leaders, community representatives and the mass media, to signal broad consensus across the wider community and involve them in the communication.

Emphasize the joint impact of taxes and expenditures

It is generally difficult to pursue a revenue-raising structural tax reform due to opposition from those who will be taxed more. Sometimes, this is due to a too narrow focus on the tax burden, without looking at the broader implications on the spending side. To convince the general public of the need and desirability of revenue-raising tax reform, it is critical to emphasize the additional expenditures they help finance. This joint impact of tax and spending can be progressive and supportive of inclusive growth, even if some of the individual taxes are regressive. Earmarking of taxes for specific expenditures should be avoided, however, since it can lead to inefficient spending decisions.

Quantify the impact of the reform

An evidence-based quantitative impact assessment is essential for several reasons. Quantification will help structure the debate and rationalize discussions among stakeholders—which might otherwise be dominated by vague statements or loose beliefs. Analysis of the impact on revenue, the income distribution and the economy will also help policy makers design the reform in the best possible manner. This enables the government to convince stakeholders that the reform is both inclusive and growth friendly. Quantitative analysis also supports the transparency and accountability of the reform process and ultimately helps build trust in government.

Use opportunities during good times

Tax reforms have been most successful when undertaken during good times, when a reduction in the overall tax burden can be used to compensate losers. For example, growth friendly tax reforms—characterized by rate reductions and base broadening—took place in the 1980s in the US and UK. Countries in continental Europe successfully moved their tax burdens in the 1990s away from direct toward indirect taxation. While growth friendly, not all these reforms were inclusive, however, and some may have increased income inequality.

Use opportunities during bad times

During or after a crisis, policymakers under pressure may rush into measures that risk damaging inclusive growth, e.g. through quick fixes like tax rate increases or the introduction of new distortive transaction taxes. However, in some countries a crisis paved the way for the introduction of long-lasting structural reforms that support inclusive growth. Indeed, crisis times may offer an opportunity for reform as the urgency facilitates political agreement among different actors. Special temporary levies on top income earners and corporate profits, for instance, have been introduced in Germany to cover the costs of unification and in Japan to finance the reconstruction efforts after the 2012 earthquake. Also the COVID crisis triggered a debate on the use of such levies to cover the increased debt—perhaps as a structural measure to increase tax progression.

Sequence reforms well

Tax reforms can be either incremental or comprehensive. Incremental reforms have the advantage that they avoid large shocks in incomes or asset prices and that people can anticipate them. For instance, reforms in the taxation of housing are often incremental to avoid large disruptions in house prices. However, incremental tax reforms might not be credible if their time span is too long as they can be hard to be sustained politically. Indeed, interest groups will have time to mobilize opposition and find ways to block those reforms. A comprehensive reform might be more difficult to achieve, however, although has the appeal of creating package deals whereby the income effects of some tax measures can be offset against others. Especially when there is space for tax relief, structural improvements might thus be achieved. Some sequencing in the reform process, even under a comprehensive reform, might still be desirable to avoid too many changes for taxpayers and tax administrations at the same time, which can impose a large burden on them.

Recognize institutional constraints

Revenue agencies responsible for the implementation of the tax system should participate in the reform process, e.g. as core members of the tax reform committee. This ensures that concerns about enforcement of a reformed tax system are recognized and accounted for in the reform strategy. For instance, some reforms may introduce excessive complexity, impose undue compliance costs, or require (third-party) information that is not yet available. Legal drafting experts are also essential for the reform process to ensure that tax laws are clear and unambiguous and that they ensure tax certainty. In some countries, decentralized fiscal powers (e.g. States, provinces or municipalities) can create obstacles to the reform process or for its implementation and their interests should also be integrated into the reform management process.

Build effective administrative capacity

Capacity constraints in tax and custom administrations are often major obstacles to revenue mobilization in developing countries. Clearly, revenue administrations should have sound management and governance arrangements and modern process/systems to manage core tax functions. And enforcement generally benefits from simple, clear, and transparent legislation (including procedural and administrative regulations), remittance and withholding regimes at adequate point of collections, taxpayer segmentation strategies, and large scale information cross-matching based on extensive use of third party information. Administrative considerations should play a key role in the design of tax reform, e.g. with major focus on simplicity, voluntary compliance mechanism and easy collection by use of withholding. Reforming the tax administration to deal with new or modified tax laws usually takes time to bear fruit, especially if they require major changes in how people work and in administrative processes. Experience is therefore that revenue effects of tax reform often occur with a time lag.